effectiveness of liraglutide and lixisenatide in the treatment of type 2 diabetes ... ·...

ORIGINAL RESEARCH

Effectiveness of Liraglutide and Lixisenatidein the Treatment of Type 2 Diabetes: Real-WorldEvidence from The Health Improvement Network(THIN) Database in the United Kingdom

Michael Feher . Gabriela Vega-Hernandez . Emina Mocevic .

Brian Buysse . Melissa Myland . Geraldine S. Power .

Lise L. Nystrup Husemoen . Joseph Kim . Daniel R. Witte

Received: December 15, 2016 / Published online: March 9, 2017� The Author(s) 2017. This article is published with open access at Springerlink.com

ABSTRACT

Introduction: The glucagon-like peptide-1receptor agonists liraglutide and lixisenatide areeffective at reducing glycated hemoglobin(HbA1c) levels in patients with type 2 diabetesmellitus (T2DM). Although liraglutide hasdemonstrated superior efficacy in head-to-head

clinical trials, real-world evidence of compara-tive effectiveness is lacking. This observationalstudy aimed to assess the effectiveness ofliraglutide versus lixisenatide in UK clinicalpractice.Methods: Electronic medical records from TheHealth Improvement Network (THIN) UK pri-mary care database were analyzed. Patients agedC18 years, diagnosed with T2DM, and pre-scribed liraglutide or lixisenatide between 01May 2013 and 31 December 2015 were includedin the study. Adjusted linear regression modelscompared the difference in mean change inHbA1c, body mass index (BMI), and systolicblood pressure (SBP) after 12-month follow-up.The proportion of patients achieving glycemiccontrol (HbA1c\6.5%,\7.0%,\7.5%); HbA1creduction [1%; and weight reduction C3%within 12 months were determined. Cox pro-portional hazards modeling was used to evalu-ate the effect of treatment on time to achievingHbA1c and weight reduction targets. Healthcareresource use (HCRU) (GP, secondary care, hos-pitalizations) was compared using analysis ofcovariance.Results: The primary outcome was assessed in579 liraglutide and 213 lixisenatide new users.Fully adjusted linear regression indicated thatliraglutide reduced HbA1c significantly morethan lixisenatide (mean treatment difference-0.30; 95% CI -0.56, -0.04; p = 0.025). Com-pared to lixisenatide, liraglutide recipients were2.5 times more likely to achieve HbA1c\6.5%

Enhanced content To view enhanced content for thisarticle go to http://www.medengine.com/Redeem/0297F060303E6D59.

M. FeherChelsea and Westminster Hospital, London, UK

G. Vega-Hernandez (&)Novo Nordisk, West Sussex, UKe-mail: [email protected]

E. Mocevic � L. L. Nystrup HusemoenNovo Nordisk, Søborg, Denmark

B. Buysse � M. Myland � G. S. Power � J. KimNEMEA Centre of Excellence for RetrospectiveStudies, QuintilesIMS, London, UK

J. KimFaculty of Epidemiology and Population Health,London School of Hygiene and Tropical Medicine,London, UK

D. R. WitteDepartment of Public Health, Aarhus University,Aarhus, Denmark

D. R. WitteDanish Diabetes Academy, Odense, Denmark

Diabetes Ther (2017) 8:417–431

DOI 10.1007/s13300-017-0241-z

http://www.medengine.com/Redeem/0297F060303E6D59

http://www.medengine.com/Redeem/0297F060303E6D59

http://crossmark.crossref.org/dialog/?doi=10.1007/s13300-017-0241-z&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1007/s13300-017-0241-z&domain=pdf

(p = 0.0002). Liraglutide users were also morelikely to achieve HbA1c \7.0% (HR 2.10;p\0.0001), \7.5% (HR 1.65; p\0.0001), and[1% HbA1c reduction (HR 1.29; p = 0.0002).BMI and SBP reductions were greater for theliraglutide group but results were not signifi-cant. HCRU was comparable between treatmentgroups.Conclusion: These results from the THIN data-base indicate that liraglutide treatment pro-vided better outcomes related to glycemiccontrol.Funding: Novo Nordisk.

Keywords: GLP-1 RA; HbA1c; Liraglutide;Lixisenatide; The Health ImprovementNetwork; THIN; Type 2 diabetes

INTRODUCTION

In the management of type 2 diabetes mellitus(T2DM), the National Institute for Health andCare Excellence (NICE) guidelines in the UKrecommend a patient-centered approach toachieving and maintaining glycemic control byindividualizing target glycated hemoglobin(HbA1c) levels [1]. Although glycemic control isconsidered on a case-by-case basis, an HbA1ctarget of 6.5% (48 mmol/mol) is recommendedfor most patients with T2DM. Following drugintensification, a secondary target of 7.0%(53 mmol/mol) is suggested and 7.5%(58 mmol/mol) is considered the threshold forfurther drug escalation [2].

Glucagon-like peptide-1 receptor agonists(GLP-1 RAs) are currently recommended asthird-line treatment in diabetes management inUK clinical practice [2]. Liraglutide once daily(Victoza, Novo Nordisk, Bagsvaerd, Denmark)and lixisenatide once daily (Lyxumia,Sanofi-Aventis Groupe, Paris, France) are twoGLP-1 RAs that have demonstrated clinicalefficacy in patients with T2DM in numerousclinical studies [3–7]. Liraglutide was approvedin the EU in 2009 and lixisenatide gainedapproval in 2013. Liraglutide and lixisenatide(as add-on to metformin) have been comparedin randomized, open-label, parallel group trials[4, 6], where liraglutide was more effective than

lixisenatide in improving glycemic control [6].However, to our knowledge, no studies havecompared the effectiveness of liraglutide tolixisenatide in patients treated in real-worldclinical practice.

This study aimed to assess the effectivenessof liraglutide in comparison to lixisenatide inadult T2DM patients treated in UK primary careby using The Health Improvement Network(THIN) database. Effectiveness was determinedby assessing the effects of these treatments onHbA1c levels, body mass index (BMI), and sys-tolic blood pressure (SBP) during the 12 monthsfollowing initiation of therapy. Healthcareresource utilization (GP, secondary care, andhospitalizations) between users of liraglutideversus lixisenatide was also investigated.

METHODS

Study Design and Data Source

This observational study of adults with T2DMutilized electronic medical record (EMR) datafrom the THIN database, a large UK primary caredata source. THIN contains anonymizedmedicalrecords for over 13 million patients, of whichover 3.5 million are currently active, represent-ing nearly 6%of theUK population. Studies havedemonstrated the validity of THINdata for use inpharmacoepidemiological studies [8–10] and itsgeneralizability to the UK in terms of demo-graphics and diabetes prevalence [11].

The period of observation was from 01 May2013 (to coincide with licensing of lixisenatidein the UK) to 31 December 2015. The index datewas the first recorded prescription issued forliraglutide or lixisenatide within the study per-iod as identified by relevant drug codes withinTHIN. Clinical effectiveness was assessed fromindex date to 12 months follow-up. BaselineHbA1c, SBP, body weight, and BMI measure-ments within the 6 months prior to index datewere included to reflect the UK National Insti-tute of Health and Care Excellence (NICE) rec-ommendation of HbA1c measurements every3–6 months [2]. Similarly, 12-month follow-upmeasurements were assessed at the 12-monthfollow-up date ±3 months. Healthcare resource

418 Diabetes Ther (2017) 8:417–431

use was assessed from index date to 12 monthspost-index. The study was reviewed andapproved by the UK Independent ScientificReview Committee (SRC) under protocol16THIN050. The analysis does not contain anynew studies with human or animal subjectsperformed by any of the authors.

Study Population

Patients were eligible for inclusion in the studypopulation if they fulfilled all of the followingcriteria: (i) at least one new prescription ofliraglutide or lixisenatide between 01 May 2013and 31 December 2015, (ii) a recorded diagnosisof T2DM at any time prior to or on index (basedon Read codes), (iii) C18 years at index, and (iv)a minimum of 6 months of medical historypre-index. Exclusion criteria were (i) diagnosisof type 1 diabetes mellitus at any time, (ii)record of gestational diabetes within 12 monthsof index date, (iii) first prescription for liraglu-tide prior to 01 May 2013, (iv) insulin pre-scription prior to index, or (v) history ofmalignancy prior to index date. Patients werefollowed from liraglutide/lixisenatide initiation(index date) until the earliest of (i) record ofinsulin prescription, (ii) transfer out of practice/death, or (iii) end of the study time period.Cohorts were constructed from the main studypopulation for each outcome of interest on thebasis of availability of follow-up measurementsin order to maximize the populations availablefor assessment.

Treatment Groups

Liraglutide prescriptions were categorized as thefollowing: all doses (any exposure to all avail-able prescribed doses [0.6 mg, 1.2 mg, 1.8 mgper day] during the study period); 1.2 mg cate-gory, average daily dose (ADD) 0.6–1.5 mg; andliraglutide 1.8 mg category, ADD [1.5–2.1 mg[12]. ADD was calculated as the sum of theprescribed number of units (pens) multiplied bythe total dosage in each unit (mg), divided bythe sum of the days covered by the prescrip-tions. For lixisenatide, the standard 20 lg doseper day was used.

RESULTS

Cohort Sizes and Baseline Characteristics

The primary outcome was the absolute changein HbA1c at 12 months from index date. Sec-ondary endpoints included (i) change in BMIand SBP at 12 months from index (Fig. 1a). Ofthe 1736 liraglutide users, 914 (52.6%) wereprescribed an average dose of 1.2 mg, 93 (5.4%)were prescribed an average dose of 1.8 mg, andfor 729 (42.0%) patients the dose was unspeci-fied. On the basis of availability of outcomemeasurements, cohort 1 for both liraglutide andlixisenatide treatment groups was divided intofive cohorts for each outcome of interest asoutlined in Fig. 1a, with cohorts 2 and 3 subdi-vided into a further three categories accordingto baseline HbA1c (Fig. 1b).

For the baseline cohort (cohort 1), the meanage of patients in the lixisenatide group(57.2 years, SD 10.7) was slightly greater thanthose in the liraglutide group (55.8 years, SD10.7). Mean weight and BMI were comparablefor both treatment groups (107.0 kg, SD 20.7;37.6 kg/m2, SD 6.6 for lixisenatide vs. 109.9 kg,SD 22.5; 38.1 kg/m2, SD 7.2 for liraglutide);however, almost 30% of all patients had missingweight data. Mean HbA1c levels at baseline werehigh in both treatment groups but slightlyhigher in patients prescribed lixisenatide(9.64%, SD 1.53 lixisenatide vs. 9.49%, SD 1.63liraglutide). Mean baseline SBP was also ele-vated but comparable between cohorts(133 mmHg, SD 14 for lixisenatide vs.134 mmHg, SD 15 for liraglutide) (Table 1).

Treatment Effects

Unadjusted analyses demonstrated reductionsin absolute change in HbA1c after 12 monthsfor both liraglutide and lixisenatide initiators incohort 2 (Fig. 2), with greater reductionsobserved in the liraglutide group (-0.93%liraglutide vs. -0.70% lixisenatide) (Fig. 2a).Using the change in estimate approach, novariables altered the treatment coefficient bymore than 10%, but all were entered into thefully adjusted model as a sensitivity analysis

Diabetes Ther (2017) 8:417–431 419

(Table 2). The fully adjusted linear regressionmodel for mean HbA1c change confirmed agreater mean change in HbA1c for the liraglu-tide group compared with the lixisenatidegroup (-0.30, 95% CI -0.56 to -0.04,p = 0.025). Linear regression analyses adjustedonly for baseline HbA1c (i.e., the baseline ornull model) demonstrated a similar result(-0.28, 95% CI -0.54 to -0.02, p = 0.038). Themagnitude of the coefficient for the sensitivity

analyses using a mixed linear regression modelwas very similar to the null model (-0.27, 95%CI -0.55 to 0.01, p = 0.058), suggesting thatclustering of patients within practice did notaffect the model.

A greater proportion of the liraglutide groupcompared to the lixisenatide group in cohort 3achieved glycemic control targets of HbA1c\6.5% (10.8% vs. 4.0%), \7.0% (24.1% vs.11.7%), \7.5% (36.0% vs. 23.3%) (Fig. 3), and

Fig. 1 a THIN study cohort analysis diagram for mainbaseline study population (cohort 1) and cohorts for eachoutcome of interest according to availability of follow-upmeasurements. b Subdivision of cohorts 2 and 3 into

categories according to baseline HbA1c. HbA1c glycatedhemoglobin, SBP systolic blood pressure

420 Diabetes Ther (2017) 8:417–431

Table 1 Baseline characteristics of patients prescribed liraglutide (all doses [0.6 mg, 1.2 mg, and 1.8 mg per day], andseparately for 1.2 mg and 1.8 mg per day) or lixisenatide (20 lg per day)

Liraglutide Lixisenatide

All doses (N5 1736) 1.2 mg (N5 914) 1.8 mg (N5 93) 20 lg (N5 578)

Age (years)

Mean (SD) 55.8 (10.7) 55.8 (10.3) 55.3 (10.7) 57.2 (10.7)

Unknown/missing (%) 0 0 0 0

Sex, n (%)

Male 951 (54.8) 511 (55.9) 57 (61.3) 295 (51.0)

Female 785 (45.2) 403 (44.1) 36 (38.7) 283 (49.0)


Smoking status, n (%)a

Current 216 (12.4) 106 (11.6) 6 (6.5) 91 (15.7)

Former 539 (31.0) 266 (29.1) 35 (37.6) 190 (32.9)

Never 612 (35.3) 330 (36.1) 32 (34.4) 207 (35.8)

Unknown/missing (%) 369 (21.3) 212 (23.2) 20 (21.5) 90 (15.6)

Weight (kg)

Mean (SD) 109.9 (22.5) 110.4 (21.1) 109.3 (24.3) 107.0 (20.7)

Unknown/missing (%) 530 (30.5) 287 (31.4) 31 (33.3) 153 (26.5)

Height (m)

Mean (SD) 1.69 (0.10) 1.69 (0.10) 1.69 (0.10) 1.69 (0.10)

Unknown/missing (%) 1166 (67.2) 649 (71.0) 61 (65.6) 380 (65.7)

BMI (kg/m2)

Mean (SD) 38.1 (7.2) 38.2 (7.1) 37.1 (7.9) 37.6 (6.6)

Unknown/missing (%) 533 (30.7) 289 (31.6) 31 (33.3) 154 (26.6)

HbA1c (%)

Mean (SD) 9.49 (1.63) 9.49 (1.54) 9.66 (1.71) 9.64 (1.53)

Unknown/missing (%) 218 (12.6) 125 (13.7) 10 (10.8) 44 (7.6)

Systolic blood pressure (mmHg)

Mean (SD) 134 (15) 134 (15) 133 (15) 133 (14)

Unknown/missing (%) 425 (24.5) 225 (24.6) 28 (30.1) 136 (23.5)

Duration of diabetes (years)

Mean (SD) 8.1 (4.9) 8.0 (4.7) 8.6 (4.9) 8.3 (4.8)


Diabetes Ther (2017) 8:417–431 421

[1% HbA1c reduction (76.0% vs. 63.6%)(Fig. 4). When stratified by diabetes duration,fewer patients tended to achieve each HbA1ctarget as duration increased (Table 3). Coxregression analysis adjusted for baseline HbA1crevealed that, compared to the lixisenatidegroup, patients in the liraglutide group were 2.5times more likely to achieve HbA1c\6.5% andapproximately twice as likely to achieve HbA1c\7.0% and \7.5% within 12 months of treat-ment (Table 4). Patients in the liraglutide groupwere statistically more likely to achieve [1%reduction in HbA1c compared to patients in thelixisenatide group (Table 4).

Unadjusted analyses indicated slightlygreater BMI and SBP reductions for liraglutideusers compared to lixisenatide users (cohort 5and 6; Fig. 2b, c). However, there was no sig-nificant difference in mean change in BMI andSBP between treatment groups (linear regressionadjusted for baseline BMI and baseline SBP,respectively) (Table 2). A slightly greater pro-portion of the liraglutide group (44.9%)achieved C3% reduction in weight comparedwith the lixisenatide group (40.8%) (cohort 5;Fig. 4). Cox regression analysis adjusted forbaseline weight indicated that patients in theliraglutide group were more likely to achieveC3% reduction in weight within 12 months

compared to the lixisenatide group; however,this difference was not statistically significant(Table 4).

The overall number of GP visits recordedwithin a year from index date was the same forthe two groups. Similar trends were seen forface-to-face visits, nurse visits, and phone callsto the GP (Table 5). There were no differencesbetween the groups regarding the mean numberof secondary care visits or hospitalizations inthe year after index date (Table 5). ANCOVAwith adjustment for baseline HbA1c and lengthof time at risk revealed no significant differencesbetween the two treatment groups.

DISCUSSION

This real-world observational study indicatedthat liraglutide-treated individuals experienced

Table 1 continued


All doses (N5 1736) 1.2 mg (N5 914) 1.8 mg (N5 93) 20 lg (N5 578)

Comorbid disease history, n (%)

Cardiovascular disease 97 (5.6) 40 (4.4) b 30 (5.2)

Hepatic disease 24 (1.4) 11 (1.2) b 14 (2.4)

Urinary tract infections 65 (3.7) 25 (2.7) b 28 (4.8)

Baseline concomitant diabetes medication, n (%)

Monotherapy 683 (39.3) – – 203 (35.1)

Dual therapy 806 (46.4) – – 299 (51.7)

Triple therapy 225 (13.0) – – 70 (12.1)

BMI body mass index, HbA1c glycated hemoglobin, SD standard deviationa Percentages were based on patients with available datab Small number suppression applied for patient numbers\6

Fig. 2 12-Month treatment effects of liraglutide (all doses[0.6 mg, 1.2 mg, and 1.8 mg per day] and separately for 1.2and 1.8 mg per day, or lixisenatide (20 lg per day) withrespect to a HbA1c (cohort 2) b body mass index (cohort5), and c systolic blood pressure (cohort 6). Mean valueswere calculated for patients with available data. BMI bodymass index, HbA1c glycated hemoglobin, SBP systolicblood pressure

c

422 Diabetes Ther (2017) 8:417–431

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0M

ean

chan

ge in

HbA

1c a

fter

12

mon

ths

of th

erap

y

HbA1c (%)

Liraglutide (all doses) (N=579) Liraglutide (1.2 mg) (N=402)

Liraglutide (1.8 mg) (N=40) Lixisenatide (20 µg) (N=213)

-1.3

-1.25

-1.2

-1.15

-1.1

-1.05

-1

Mea

n ch

ange

in B

MI (

Kg/

m2)

aft

er

12 m

onth

s of

ther

apy

BMI (Kg/m2)



B

-3

-2.5

-2

-1.5

-1

-0.5

0

Mea

n ch

ange

in S

BP

(mm

Hg)

aft

er

12 m

onth

s of

ther

apy

SBP (mmHg)



C

A

Diabetes Ther (2017) 8:417–431 423

Table 2 Linear regression analyses of mean change in HbA1c, BMI, and SBP from baseline to 12 months in patientsprescribed liraglutide (all doses [0.6 mg, 1.2 mg, and 1.8 mg per day], and separately for 1.2 mg and 1.8 mg per day) versuslixisenatide (20 lg per day)

Linear regression analyses

b 95% LCL 95% UCL p value

Mean change in HbA1c: baseline modela

Lixisenatide (20 lg) –

Liraglutide (all doses) -0.278 -0.541 -0.016 0.038

Liraglutide (1.2 mg) -0.362 -0.633 -0.092 0.009

Liraglutide (1.8 mg) -0.417 -0.965 0.130 0.135

Mean change in HbA1c: fully adjusted modelb


Liraglutide (all doses) -0.297 -0.556 -0.037 0.025

Liraglutide (1.2 mg) -0.395 -0.663 -0.128 0.004

Liraglutide (1.8 mg) -0.502 -1.040 0.036 0.068

Mean change in BMI: baseline modelc


Liraglutide (all doses) -0.009 -0.488 0.469 0.969

Liraglutide (1.2 mg) -0.005 -0.526 0.516 0.985

Liraglutide (1.8 mg) -0.076 -1.177 1.026 0.893

Mean change in SBP: baseline modeld


Liraglutide (all doses) -0.404 -2.644 1.837 0.724

Liraglutide (1.2 mg) -0.851 -3.082 1.381 0.454

Liraglutide (1.8 mg) -0.171 -4.741 4.399 0.941

b b-coefficient, BMI body mass index, HbA1c glycated hemoglobin, LCL lower confidence limit, SBP systolic bloodpressure, UCL upper confidence limita Linear regression adjusted for treatment and baseline HbA1c only. N = 784 for the model with all doses of liraglutide andN = 649 for the model with doses split into 1.2 and 1.8 mgb Linear regression adjusted for all covariates (for baseline HbA1c, age, gender, smoking status, history of hepatic disease,history of urinary tract infection, concomitant medication, history of cardiovascular disease, and diabetes duration). BMIwas not adjusted for because of a high level of missing values and an assumed assumption of causal linkage between exposureand outcome. N = 784 for the model with all doses of liraglutide and N = 649 for the model with doses split into 1.2 and1.8 mgc Linear regression adjusted for treatment and baseline BMI only. N = 569 for the model with all doses of liraglutide andN = 478 for the model with doses split into 1.2 and 1.8 mgd Linear regression adjusted for treatment and baseline SBP only. N = 666 for the model with all doses of liraglutide andN = 559 for the model with doses split into 1.2 and 1.8 mg

424 Diabetes Ther (2017) 8:417–431

greater reductions in HbA1c than lixisen-atide-treated individuals during the 12 monthsfollowing treatment initiation. Additionally, a

greater proportion of patients initiatingliraglutide achieved HbA1c targets compared tolixisenatide patients. These findings from

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

<6.5% HbA1c target <7.0% HbA1c target <7.5% HbA1c target

Prop

ortio

n of

Pat

ient

s

Liraglutide (all doses) Liraglutide (1.2 mg) Liraglutide (1.8 mg) Lixisenatide (20 µg)

Fig. 3 Proportion of patients in cohort 2 achievingHbA1c levels \6.5%, \7.0%, and \7.5% within12 months of therapy with liraglutide (all doses [0.6 mg,1.2 mg, and 1.8 mg per day] and separately for 1.2 mg and

1.8 mg per day) or lixisenatide (20 lg per day). Propor-tions were calculated on the basis of patients with availabledata. HbA1c glycated hemoglobin

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

>1% reduction in HbA1c ≥3% weight reduction

Prop

ortio

n of

Pat

ient

s

Liraglutide (all doses) Liraglutide (1.2 mg) Liraglutide (1.8 mg) Lixisenatide (20 µg)

Fig. 4 Proportion of patients achieving [1% HbA1creduction (cohort 3) and C3.0% reduction in weight(cohort 4) within 12 months of therapy initiation withliraglutide (all doses [0.6 mg, 1.2 mg, and 1.8 mg per day],

and separately for 1.2 mg and 1.8 mg per day) orlixisenatide (20 lg per day). Proportions were calculatedon the basis of patients with available data. HbA1c glycatedhemoglobin

Diabetes Ther (2017) 8:417–431 425

real-world clinical practice are aligned withresults from head-to head randomized con-trolled trials (RCT) where patients in liraglutidetreatment groups demonstrated greater reduc-tions in HbA1c levels than patients in lixisen-atide treatment groups [4–6].

Baseline clinical characteristics were similarbetween treatment groups. Approximately threetimes more patients were prescribed liraglutidecompared to lixisenatide, which may be due to

longer market availability and thus increasedfamiliarity with liraglutide. Importantly, thepresent study found that patients prescribedliraglutide or lixisenatide had higher meanbaseline HbA1c levels compared to patients inRCTs, but this could be influenced by inclusioncriteria thresholds for RCTs [4–6]. In clinicalpractice, high mean HbA1c could be explainedby current UK NICE guidelines recommendingGLP-1 RA as a third-line treatment option in

Table 3 Proportion of patients prescribed liraglutide (all doses [0.6 mg, 1.2 mg, and 1.8 mg per day], and separately for1.2 mg and 1.8 mg per day) or lixisenatide (20 lg per day) achieving HbA1c and weight reduction targets within 12 monthsof therapy, stratified by diabetes duration


All doses% (n/N)

1.2 mg% (n/N)

1.8 mg% (n/N)

20 lg% (n/N)

HbA1c target\6.5%

Diabetes duration\5 years 16.9 (63/372) 21.1 (47/223) a 11.2 (14/125)

Diabetes duration 5–10 years 9.9 (43/436) 13.6 (38/279) a a

Diabetes duration[10 years 6.2 (25/405) 7.2 (17/235) a a

HbA1c target\7.0%

Diabetes duration\5 years 30.6 (110/360) 37.9 (83/219) 33.3 (7/21) 22.8 (28/123)

Diabetes duration 5–10 years 23.7 (102/431) 29.8 (82/275) 25.9 (7/27) 9.7 (16/165)

Diabetes duration[10 years 18.7 (74/395) 24.6 (57/232) 22.2 (6/27) 5.1 (8/158)

HbA1c target\7.5%




[1% HbA1c reduction




C3% weight reduction

Diabetes duration\5 years 49.0 (51/104) 58.7 (37/63) a 51.6 (16/31)

Diabetes duration 5–10 years 37.6 (38/101) 45.2 (28/62) a 40.8 (20/49)


HbA1c glycated hemoglobina Small number suppression applied for patient numbers\6

426 Diabetes Ther (2017) 8:417–431

patients for whom triple therapy with met-formin and two other oral drugs was either noteffective, not tolerated, or contraindicated [2].

When assessing the primary outcome,notable differences between the treatmentgroups were observed for patients with12 months of available data where patients ini-tiating liraglutide achieved a significantlygreater reduction in HbA1c than those initiatedon lixisenatide. This supports previous head-to-head RCTs where liraglutide demonstratedsuperior reductions in HbA1c levels comparedwith lixisenatide in 26-week [6], 28-day [4], and8-week RCTs [5].

Additionally, results from an observationalstudy suggest that the observed greater mean

reduction in HbA1c may be regarded as clinicallyrelevant [13]. The findings indicated that reduc-ing HbA1c by 1% may reduce the risk of dia-betes-related death by 21% and all-causemortality by 14%. These results, however, couldnot be reproduced in the three major glu-cose-lowering trials ADVANCE, ACCORD, andVADT [14–16].More recently,meta-analyses havefound that reducing HbA1c could significantlyreduce the incidence of non-fatal myocardialinfarction and coronary heart disease, althoughthis is less certain for all-cause mortality [17, 18].

Approximately three-quarters of the liraglu-tide group and two-thirds of the lixisenatidegroup achieved[1% reduction in HbA1c. Theseproportions were greater than observed in two

Table 4 Cox proportional hazards (PH) regression analyses for liraglutide (all doses [0.6 mg, 1.2 mg, and 1.8 mg per day])versus lixisenatide (20 lg per day) to assess the effect of exposure on time to achieving HbA1c and weight reduction targetswithin 12 months of therapy

Cox PH regression analyses

HR 95% LCL 95% UCL p value

Probability of achieving HbA1c\6.5%a

Lixisenatide (20 lg) (N = 454) –

Liraglutide (all doses) (N = 1213) 2.54 1.55 4.16 0.0002



Liraglutide (all doses) (N = 1186) 2.10 1.57 2.83 \0.0001



Liraglutide (all doses) (N = 1134) 1.65 1.33 2.05 \0.0001

Probability of achieving[1% HbA1c reductiona



Probability of achieving C3% weight reductionb



HbA1c glycated hemoglobin, HR hazard ratio, LCL lower confidence limit, PH proportional hazards, UCL upper confi-dence limita Cox PH regression adjusted for baseline HbA1c onlyb Cox PH regression adjusted for baseline weight only

Diabetes Ther (2017) 8:417–431 427

other real-world liraglutide studies, where justover half of liraglutide users achieved C1%HbA1c reduction after 6 months of treatment[19, 20]. Furthermore, patients initiated onliraglutide had a statistically significant higherprobability than lixisenatide users of achievingglycemic targets (including \6.5%) over12 months. We found that glycemic control

targets were attained by a greater proportion ofthe liraglutide group versus the lixisenatidegroup. This largely agrees with RCT and real--world data, although the relative proportions ofpatients achieving these targets were lower thanin clinical trials examining liraglutide 1.8 mg[6, 21]. These lower proportions could beexplained by the high mean baseline HbA1c in

Table 5 GP, secondary care visits, and hospitalizations of liraglutide users (all doses [0.6 mg, 1.2 mg, and 1.8 mg]) andlixisenatide users during 12 months following initiation of treatment

All patients Liraglutide Lixisenatide

All GP visits

Number of patients 2247 1681 566

Mean (SD) 9.8 (7.5) 9.7 (7.5) 9.8 (7.4)

Median (IQR) 8.0 (5.0, 13.0) 8.0 (5.0, 13.0) 8.0 (5.0, 13.0)

Face-to-face GP visits


Mean (SD) 7.2 (5.6) 7.2 (5.6) 7.2 (5.4)

Median (IQR) 6.0 (3.0, 10.0) 6.0 (3.0, 10.0) 6.0 (3.0, 9.0)

GP nurse visits


Mean (SD) 3.0 (3.0) 3.0 (2.8) 3.1 (3.3)

Median (IQR) 2.0 (1.0, 4.0) 2.0 (1.0, 4.0) 2.0 (1.0, 4.0)

GP calls


Mean (SD) 2.4 (2.3) 2.4 (2.3) 2.4 (2.4)

Median (IQR) 2.0 (1.0, 3.0) 2.0 (1.0, 3.0) 2.0 (1.0, 3.0)

Secondary care visits


Mean (SD) 2.4 (1.4) 2.4 (1.5) 2.4 (1.4)

Median (IQR) 2.0 (1.0, 3.0) 2.0 (1.0, 3.0) 2.0 (1.0, 3.0)

Hospitalizations


Mean (SD) 1.3 (0.7) 1.3 (0.7) 1.3 (0.6)

Median (IQR) 1.0 (1.0, 1.0) 1.0 (1.0, 1.0) 1.0 (1.0, 1.0)

IQR interquartile range, SD standard deviation

428 Diabetes Ther (2017) 8:417–431

our study, which has previously been identifiedas a significant predictor of achieving HbA1ctargets in response to liraglutide [22]. Responsemay also be influenced by duration of diabetes;our results indicated a poorer response to theglycemic reduction and control targets withincreased duration, a finding consistent withthe Association of British Clinical DiabetologistsNationwide Liraglutide Audit [23].

Despite greater improvement in glycemiccontrol with liraglutide use, both liraglutideand lixisenatide use was associated with a sim-ilar number of GP, secondary care, and hospitalvisits in the year following index. Results from along-term clinical trial have suggested thattreatment with liraglutide could be associatedwith reduced resource use due to a reduction inmicrovascular and macrovascular events [24].

Our study supports the findings of RCTs andobservational studies but carries potential limi-tations. For some outcomes (i.e., HbA1c,weight), data were not available at both baselineand 12-month follow-up, which could lead toselection bias. However, time windows formeasurements were decided according toguidelines for measurement frequency [2] inorder to maximize the number of patients in thestudy. Confounding by indication (i.e., indica-tion bias) resulting from differences in reasonsfor prescription may lead to systematic bias infavor or against liraglutide. Indication bias maybe limited since both liraglutide and lixisen-atide belong to the same drug class and both areused as a third-line treatment in the UK. How-ever, the longer market availability and thusfamiliarity of liraglutide may result in it beingpreferentially prescribed, as was reflected in thegreater number of patients prescribed liraglu-tide. Multiple linear regression models wereapplied to mitigate any possible limitations interms of confounding. The current study wasalso limited by data contained within the THINdatabase. Firstly, medication data are based onprescriptions issued rather than prescriptionsdispensed and thus it was not possible to con-firm that the drug was actually taken by thepatient. Secondly, GPs in England receivedincentives during the study time period forreporting diabetes, while GPs in Wales, North-ern Ireland, and Scotland did not. This may

have led to an oversampling of patients fromEnglish GPs and a higher quality of recordingcompared to GPs from Wales, Northern Ireland,and Scotland.

CONCLUSIONS

These data from UK clinical practice providereal-world evidence that liraglutide comparedto lixisenatide is associated with greater reduc-tions in HbA1c and a greater likelihood ofachieving HbA1c targets, supporting the resultsfrom previous clinical trials.

ACKNOWLEDGEMENTS

Sponsorship for this study and article process-ing charges were funded by Novo Nordisk. Allnamed authors meet the International Com-mittee of Medical Journal Editors (ICMJE) cri-teria for authorship for this manuscript, takeresponsibility for the integrity of the work as awhole, and have given final approval for theversion to be published. All authors had fullaccess to all of the data in this study and takecomplete responsibility for the integrity of thedata and accuracy of the data analysis. Writingassistance in the preparation of this manuscriptwas provided by Dr. Heidi Eriksen of AandooLtd. Support for this assistance was funded byQuintilesIMS.

Disclosures. M. Feher has received financialsupport for speaker meeting/consultancy/re-search from Novo Nordisk and Sanofi.G. Vega-Hernandez, E. Mocevic, B. Buysse, M.Myland, S. Power, L. L. Nystrup Husemoen, J.Kim, and D. R. Witte have nothing to disclose.

Compliance with Ethics Guidelines. Thisstudy was reviewed and approved by the UKSRC under protocol 16THIN050 and was per-formed in accordance with the ethical standardsof the 1964 Declaration of Helsinki. This anal-ysis does not contain any new studies withhuman or animal subjects performed by any ofthe authors.

Diabetes Ther (2017) 8:417–431 429

Open Access. This article is distributedunder the terms of the Creative CommonsAttribution-NonCommercial 4.0 InternationalLicense (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommer-cial use, distribution, and reproduction in anymedium, provided you give appropriate creditto the original author(s) and the source, providea link to the Creative Commons license, andindicate if changes were made.

REFERENCES

1. McGuire H, Longson D, Adler A, Farmer A, Lewin I.Management of type 2 diabetes in adults: summaryof updated NICE guidance. BMJ. 2016;353:i1575.

2. National Institute of Health and Care Excellence.Type 2 diabetes in adults: management, NICEguidelines [NG28]. https://www.nice.org.uk/guidance/NG28/chapter/1. Accessed August 2016.

3. Bain SC. The clinical development program oflixisenatide: a once-daily glucagon-like peptide-1receptor agonist. Diabetes Ther. 2014;5(2):367–83.

4. Kapitza C, Forst T, Coester HV, Poitiers F, Ruus P,Hincelin-Mery A. Pharmacodynamic characteristicsof lixisenatide once daily versus liraglutide oncedaily in patients with type 2 diabetes insufficientlycontrolled on metformin. Diabetes Obes Metab.2013;15(7):642–9.

5. Meier JJ, Rosenstock J, Hincelin-Mery A, Roy-DuvalC, Delfolie A, Coester HV, et al. Contrasting effectsof lixisenatide and liraglutide on postprandial gly-cemic control, gastric emptying, and safety param-eters in patients with type 2 diabetes on optimizedinsulin glargine with or without metformin: a ran-domized, open-label trial. Diabetes Care.2015;38(7):1263–73.

6. Nauck M, Rizzo M, Johnson A, Bosch-Traberg H,Madsen J, Cariou B. Once-daily liraglutide versuslixisenatide as add-on to metformin in type 2 dia-betes: a 26-week randomized controlled clinicaltrial. Diabetes Care. 2016;39(9):1501–9.

7. Sethi BK, Viswanathan V, Kumar A, Chatterjee S,Chandalia HB. Liraglutide in clinical practice:insights from LEAD programme. Suppl JAPI.2010;58(June):18–22.

8. Lewis JD, Schinnar R, Bilker WB, Wang X, Strom BL.Validation studies of The Health Improvement Net-work (THIN) database for pharmacoepidemiology

research. Pharmacoepidemiol Drug Saf.2007;16(4):393–401.

9. Denburg MR, Haynes K, Shults J, Lewis JD, LeonardMB. Validation of The Health Improvement Net-work (THIN) database for epidemiologic studies ofchronic kidney disease. Pharmacoepidemiol DrugSaf. 2011;20(11):1138–49.

10. Martin-Merino E, Fortuny J, Rivero E, Garcia-Ro-driguez LA. Validation of diabetic retinopathy andmaculopathy diagnoses recorded in a UK primarycare database. Diabetes Care. 2012;35(4):762–7.

11. Blak BT, Thompson M, Dattani H, Bourke A. Gen-eralisability of The Health Improvement Network(THIN) database: demographics, chronic diseaseprevalence and mortality rates. Inform Prim Care.2011;19(4):251–5.

12. Divino V, DeKoven M, Hallinan S, et al. Glucagon--like peptide-1 receptor agonist treatment patternsamong type 2 diabetes patients in six Europeancountries. Diabetes Ther. 2014;5(2):499–520.

13. Stratton IM, Adler AI, Neil HA, et al. Association ofglycaemia with macrovascular and microvascularcomplications of type 2 diabetes (UKPDS 35):prospective observational study. BMJ.2000;321(7258):405–12.

14. Duckworth W, Abraira C, Moritz T, et al. Glucosecontrol and vascular complications in veterans withtype 2 diabetes. N Engl J Med. 2009;360(2):129–39.

15. Heller SR. A summary of the ADVANCE Trial. Dia-betes Care. 2009;32(Suppl 2):S357–61.

16. The Action to Control Cardiovascular Risk in Dia-betes Study Group. Effects of intensive glucoselowering in type 2 diabetes. N Engl J Med.2008;358(24):2545–59.

17. Ray KK, Seshasai SR, Wijesuriya S, et al. Effect ofintensive control of glucose on cardiovascular out-comes and death in patients with diabetes mellitus:a meta-analysis of randomised controlled trials.Lancet. 2009;373(9677):1765–72.

18. Turnbull FM, Abraira C, Anderson RJ, et al. Inten-sive glucose control and macrovascular outcomesin type 2 diabetes. Diabetologia.2009;52(11):2288–98.

19. Lind M, Matsson PO, Linder R, et al. Clinicaleffectiveness of liraglutide vs sitagliptin on gly-cemic control and body weight in patients withtype 2 diabetes: a retrospective assessment in Swe-den. Diabetes Ther. 2016;7(2):321–33.

20. Nyeland ME, Ploug UJ, Richards A, et al. Evaluationof the effectiveness of liraglutide and sitagliptin in

430 Diabetes Ther (2017) 8:417–431

http://creativecommons.org/licenses/by-nc/4.0/

http://creativecommons.org/licenses/by-nc/4.0/

https://www.nice.org.uk/guidance/NG28/chapter/1

https://www.nice.org.uk/guidance/NG28/chapter/1

type 2 diabetes: a retrospective study in UK primarycare. Int J Clin Pract. 2015;69(3):281–91.

21. Ostawal A, Mocevic E, Kragh N, Xu W. Clinicaleffectiveness of liraglutide in type 2 diabetes treat-ment in the real-world setting: a systematic litera-ture review. Diabetes Ther. 2016;7(3):411–38.

22. Aroda VR, Brett J, Khutoryansky N, Ratner R.Identifying predictors of response to liraglutide intype 2 diabetes using recursive partitioning analy-sis. Can J Diabetes. 2012;36(5):S45.

23. Thong KY, McGowan BY, Htay T, Pernet A, Kelly C,Rajeswaran C, et al. Insulin treatment and longerdiabetes duration both predict poorer glycaemicresponse to liraglutide treatment in type 2 diabetes:the Association of British Clinical DiabetologistsNationwide Liraglutide Audit. Br J Diabetes VascDis. 2016;15(4):169–72.

24. Marso SP, Daniels GH, Brown-Frandsen K, et al.Liraglutide and cardiovascular outcomes in type 2diabetes. N Engl J Med. 2016;375(4):311–22.

Diabetes Ther (2017) 8:417–431 431

effectiveness of liraglutide and lixisenatide in the treatment of type 2 diabetes ... ·...

Documents